This is a writeup of a basic analysis of a slice of data from "Metastudies for robust tests of theory." by Baribault et al.
The experimental task consisted of subjects, over multiple sessions, clicking an indicated target either quickly or accurately, depending on which cue they recieved. The cues were either visible or masked. Performance at this task nominally reveals whether the value of the masked cue, while not consciously perceptible, impacted participant behavior.
The analytic objective is to:
- assess the effect size of cue condition (visibility x instruction)
- for each experimental session, use linear regression to find the intercepts and coefficients for the Instruction variable in each visibility condition
- repeat these regressions with an additional penalty term to bias these estimates towards the grand mean of their values across all sessions.